The Impact of HIV/AIDS on the South African Economy: A Review of ...

The Impact of HIV/AIDS on the South
African Economy: A Review of Current
Evidence
Booysen, F. le R., Geldenhuys, J.P. & Marinkov, M.
Department of Economics
University of the Free State
THE CHALLENGE OF GROWTH AND POVERTY: THE SOUTH AFRICAN ECONOMY SINCE DEMOCRACY.
T
I
P
S
/
D
P
R
U

F
o
r
u
m

2
0
0
3

---

2
The Impact of HIV/AIDS on the South African Economy: A Review of
Current Evidence1

Booysen, F. le R., Geldenhuys, J.P. & Marinkov, M.
Department of Economics
University of the Free State
Bloemfontein, South Africa

Paper prepared for TIPS/DPRU conference on ‘The Challenge of Growth and Poverty: The
South African economy since democracy’
8-10 September 2003, Indaba Hotel, Johannesburg

Abstract

Even though the approaches, assumptions and results may vary greatly in the macroeconomic
models employed in estimating the impact of HIV/AIDS on the South African economy, the
overriding message that these models convey remains the same: the cost of HIV/AIDS to South
Africa will be significant in economic, social and human terms. However, the accuracy of the models
and their results can be faulted for various reasons, not least the shortcomings of current
demographic projections and the empirical evidence on the microeconomic impact of the epidemic,
shortcomings that can be argued to translate into both under- and overestimation of the likely
macroeconomic impacts of the epidemic. More work is also required to quantify the nature of the
impact of the epidemic on specific sectors in the economy. In addition, more recent, alternative
methodological approaches can also be explored in further investigating the macroeconomic
impacts of the epidemic. Finally, models are also constrained by a lack of clarity regarding the key
question of how treatment, care and support for HIV/AIDS-affected individuals and households are
to be financed in South Africa, given that government at times are unclear as to what policies will be
implemented to fights HIV/AIDS.

South Africa currently faces one of the highest HIV prevalence rates in the world. The estimated
adult prevalence of HIV amongst 15-49 year olds in 2001 was 20.1% (UNAIDS, 2002), while the
ASSA2000 model put adult prevalence amongst 20-65 year olds (in the unchanged scenario) at
24.1% (ASSA, 2003). A recent national household survey in turn has put the 2002 estimate of adult

1 This research paper has been prepared with the aid of financial support from Trade and Industrial Policy
Strategies (TIPS).

---

3
prevalence amongst those older than 25 years at 15.5% (HSRC, 2002)2. Given that HIV/AIDS
primarily effects the economically and sexually active population, the epidemic poses a serious
threat to economic growth, development prospects and poverty alleviation. In fact, the predicted
macroeconomic impacts of the HIV/AIDS epidemic make light of the macroeconomic targets of
GEAR, given the projected decline in economic growth and employment.

The main aim of this paper is to review the current literature and evidence of the impact of
HIV/AIDS on the South African economy. The paper is structured as follows. Section 1 provides a
brief overview of the methodology of the four macroeconomic models employed in estimating the
impacts of the epidemic that are reviewed in this paper. (It should be emphasized however that this
is not a methodological review of macroeconomic modeling, which is outside the scope of this
paper.) Given that these models project the macroeconomic impacts of the HIV/AIDS epidemic over
a 10-15 year period that ranges from 2000 to 2015 and that the HIV epidemic is yet to evolve into a
full-scale AIDS epidemic, the emphasis in this paper is therefore on the future challenges that
HIV/AIDS poses to the South African economy, rather than the challenges during the first 10 years
of democracy3. Section 2 describes the main economic impact channels of the HIV/AIDS epidemic
as described in these four macroeconomic models, whilst section 3 and 4 respectively focus on an
overview of the assumptions (input) and projected impacts on economic growth, investment,
employment, and poverty (outputs) of these four models. The assumptions and projections of these
models are critically adjudged at the hand of currently available empirical evidence on the
economics of HIV/AIDS in South Africa. In section 5, the implications to the macroeconomic
modelling results of recent changes in the responses of government, business, communities and
other role players in South Africa to the HIV/AIDS epidemic are discussed. Section 6 concludes,
summarizing the main lessons to be learned from the review and the key questions that remain
unanswered by current research on the economics of HIV/AIDS in South Africa.

1.
Macroeconomic modelling of the impact of HIV/AIDS

Initially, the primary focus in HIV/AIDS modelling was demographic, behavioural and
epidemiological rather than economic in nature. However, some models have been developed and
employed in estimating the resource requirements for financing prevention, care and treatment, and
support interventions aimed at curbing the spread of the epidemic and mitigating its adverse

2 The fact that these estimates are based on prevalence by different age categories precludes a direct
comparison of these specific estimates of HIV prevalence.
3 However, it should be pointed out that Whiteside and Sunter (2000) argue that government’s responses to
the epidemic have lacked urgency and focus and that the situation could have been different had this not
been the case (they point out that the HIV prevalence rate in 1994 stood at 7.6% only, a figure that has since
escalated to twice this figure). Section 6 of this paper again touches on these policy shifts with regard to
HIV/AIDS over the past 10 years.

---

4
impacts on society, notably the Resource Needs Model (RNM), Goals Model (GM) and cost
modules of the Spectrum model (Van der Heever, 2003). In more recent times, though, several
models have been employed in directly modelling the future macroeconomic impacts of the
HIV/AIDS epidemic on the South African economy. Ford et al. (2002) distinguishes these models
from so-called broad qualitative evaluations of the impact of HIV/AIDS on macroeconomic variables
and the case study approach, which entails the application of lessons from other country studies to
say South Africa. According to Ford et al. (2002), current research on the macroeconomic impact of
the epidemic ‘seek, in general, to quantify the effect of the epidemic as an endogenous shock on a
volatile, emerging and globalised economic system’. Four such models are reviewed in this paper,
i.e., the Arndt and Lewis model (2000), the ING Barings model (2000), Burger’s (2001) model, and
the BER (2001) model4. These models each follow a different methodological approach to
modelling the economic impact of HIV/AIDS. The ING Barings (2000) and BER (2001) models
follow a demand-side driven approach, while Burger (2001) follows a supply-side driven approach.
Arndt and Lewis (2000) employ a supply-constrained Computable General Equilibrium (CGE) model
in estimating the macroeconomic impact of HIV/AIDS.

Macroeconomic models of the economic impact of HIV/AIDS all require demographic inputs, i.e.
actuary-based estimates of the impact of the HIV/AIDS epidemic on the size and structure of the
population. To date, most of these models have employed either the earlier Doyle-Metropolitan
demographic projections or the later projections from the ASSA2000 model.

Arndt and Lewis (2000) employ an economy-wide supply-constrained CGE model, the focus being
South African economy’s medium-term growth prospects. The ING Barings (2000) model employs
the WEFA consultancy group’s annual macro-econometric framework, which is based on the
principle that in the short run, demand factors will dominate the economy, whereas in the long run,
supply factors are more dominant. The ING Barings (2000) model therefore takes both the supply
and the demand factors into account in the econometric modelling and, unlike the Arndt and Lewis
(2000) model, the economic forecasts are mainly of a long-term nature.

4 This review excludes three other studies attempting to estimate the economic impact of HIV/AIDS. Rather
than estimate the various macroeconomic impacts of the HIV/AIDS epidemic, Broomberg et al. (1991)
employed the human capital approach to estimate the impact of HIV/AIDS in South Africa. The human capital
approach entails the use of lost earnings as a proxy for total lost production attributable to HIV/AIDS. Haacker
(2002) in turn discuss the macroeconomic impact of the HIV/AIDS epidemic and applies a supply-side Cobb-
Douglas approach in modelling the impact of the HIV/AIDS epidemic. However, Haacker (2002) discusses
this modelling approach in general and without any reference to or direct application to the South African
economy. Finally, the recently published report on the estimated intergenerational impact of HIV/AIDS on the
South African economy (Bell et al., 2003), which although theoretically sound it may be said employs
questionable assumptions in some instances with regard to demographic and economic responses to the
epidemic, is not discussed here insofar as it employs a forecasting horizon (80 years) beyond those of the
models discussed in this paper and does not report comparable forecasts for similar macroeconomic impacts
of the epidemic to those discussed in this particular paper.

---

5

The BER (2001) model also makes use of econometric modelling techniques. BER (2001) stresses
that their study should be seen as a “macroeconomic sensitivity analysis” rather than a forecasting
model, with the baseline scenario reflecting the “worst case demographic scenario” that does not
allow for behavioural changes or large-scale government intervention.

Supply-side modelling is also a popular approach to determine the impacts of HIV/AIDS on the
economy. This type of modelling is fairly simple and less sophisticated than other approaches.
Burger (2001) extends the simple Cobb-Douglas production function to incorporate the key
macroeconomic variables affected by HIV/AIDS. Being a supply-side oriented model, it relies solely
on the factors of production and is used to calculate potential output, i.e. the level of aggregate
output that can be sustained in the long run with stable inflation. The essence of this model is that
the estimate of the long run growth rate of real GDP is represented as the sum of the growth rates
of the labour force, capital and technology.

2. The Main Channels of the Economic Impact of HIV/AIDS

HIV/AIDS affects the economy on the micro- as well as the macro-levels. From a macroeconomic
perspective, HIV/AIDS would amongst others affect labour supply and demand, capital investment
and utilisation, savings, investment, aggregate demand, GDP and the distribution of income. These
macroeconomic impacts follow from a range of micro-level impacts, including the impact of the
HIV/AIDS epidemic on business, economic sectors, the public sector, and households (Bollinger
and Stover, 1999). The four macroeconomic models reviewed in this paper focus on five primary
impact channels (and which drives the assumptions included in these models). These impact
channels are the following:

-
A decline in total labour supply and in the total population due to HIV/AIDS-related mortality
amongst the economically active population, which affects both the demand and supply side of
the economy.

-
A decline in labour and total factor productivity resulting from HIV/AIDS-related morbidity.

-
Direct and indirect costs and productivity losses to the private sector due to HIV/AIDS-related
morbidity and mortality amongst employees: Firms will experience higher expenditure due to
increased health care costs, burial fees and training costs and payment of other employee
benefits, as well as absenteeism and a higher labour turnover, which will increase the cost of

---

6
employment. In addition, demand for certain goods and service and therefore sales revenue
and turnover may increase or decline due to changes in household expenditure patterns
(Bollinger and Stover, 1999). These impacts of the epidemic translate into increased production
costs, increased prices (i.e. higher PPI, and perhaps even higher interest rates), and a decline
in aggregate demand, savings and investment.

-
Household expenditure: The care of and loss of HIV infected family members translate into
losses of household income as well as higher medical and funeral expenses, which results in
changes in expenditure patterns and in turn in private savings and in investment (Bollinger and
Stover, 1999).

-
Government expenditure: HIV/AIDS will impact primarily on the health sector due to a higher
demand for health services and the high costs of HIV/AIDS treatment. In addition, the public
sector will like business in the private sector face higher costs of employment and lower
productivity as a result of HIV/AIDS (Bollinger and Stover, 1999), which in turn will result in
lower savings due to greater deficits.

3.
Model Assumptions

The assumptions of the different models are first presented in tabular form for the sake of simplicity,
using the five main impact channels to structure this discussion, after which these assumptions are
summarised and evaluated critically with reference to the existing empirical evidence of the
economic impact of HIV/AIDS in South Africa. At the end of each section, the main reasons why the
overall (or specific) macroeconomic impacts of the epidemic may be over- or underestimated are
presented in brief.

Table 1: Model assumptions
Impact Channel
Arndt and Lewis
ING Barings (2000)
BER
Burger

(2000)
(2001)
(2001)

---

7
A. Population
-
ASSA demographic
-
ASSA demographic
-
Doyle-Metropolitan
-
ASSA2000
and labour
projections.
projections.
and ASSA2000
demographic
supply
-
HIV/AIDS will cause
-
Labour supply will be
demographic
projections.

a slower growth in
lower due to
projections.
-
Individuals infected
population and
HIV/AIDS.
-
Lower fertility
with HIV have a life
labour supply by skill
-
For every person
together with AIDS
expectancy of 10
category.
with full-blown AIDS,
deaths will cause a
years and are
-
Effective labour input
four months of
lower population and
expected to live 8
for each skill type will
“person-year
a lower labour force.
years before falling ill
be reduced
equivalent” labour
-
Model differentiates
with AIDS.
proportionally with
supply will be lost.
between the
-
Infection rate is
projected AIDS
-
Average wage is
medium- and high-
derived from the
deaths one period
used as a
skilled labour and the
number of projected
hence.
reasonable proxy for
semi- and unskilled
AIDS deaths.

skills; highly skilled
labour force since

labour earn 1.91
the latter can be
times more than the
replaced at a lower
“economy-wide
cost while having a
average”; skilled
prevalence rate that
labour earn 1.07
is much higher
times more than the
compared to
“economy-wide
medium- and high-
average”; semi
skilled labour.
skilled and unskilled

labour earn 0.65
times the “economy -
wide average”.
B. Labour and
-
AIDS afflicted
-
Increased
-
The productivity of
-
Productivity is only
total factor
workers stay on the
absenteeism due to
skilled and unskilled
allowed to vary with
productivity
workforce for two
HIV/AIDS will result
workers infected with
age and population
years.
in a decline in labour
HIV/AIDS will be
groups.
-
Incidence of
productivity.
reduced by 40%.
-
Determinants of
HIV/AIDS among
-
Productivity loss
-
Total factor
labour productivity
workers will reduce
(uniform to all skill
productivity growth is
are experience and
labour productivity
categories) is four
21% lower in the
to some degree
AIDS afflicted
months per annum.
AIDS scenario – this
schooling.
workers are half as
is based on the 21%
-
Assumption is that
productive as the
reduction in the total
experience is equal
rest of the labour
labour force due to
to workers’ age less
force.
AIDS.
fifteen years; this
-
Total factor
allows one to see the
productivity is lower
composition as well
due to prevalence of
as the experience of
HIV/AIDS.
the labour force.
-
At the height of the
epidemic, total factor
production growth
will fall to one half of
the no-AIDS rate.

---

8
C. Direct and
-
Insurance and
-
Companies face
-
Semi-skilled and
-
Half of the direct
indirect costs
benefit payments will
higher direct costs
unskilled labour was
costs of HIV/AIDS is
for the private
increase, thus putting
and indirect costs.
excluded from the
absorbed by
sector
pressure on costs,
-
Direct costs of skilled
direct cost estimates,
companies and the
profits and savings.
and highly skilled
given that these
other half is passed
-
Higher costs together
employees would
employees are
on to employees.
with absenteeism
increase from 15% in
unlikely to be
and labour turnover
2005 to 30% in 2010.
covered by medical
may result in a
-
Indirect costs of
aid.
higher capital
skilled and highly
-
20% of South
intensity of firms.
skilled employees
Africans are covered
-
Transaction costs in
would increase from
by employment-
enforcing of
10% in 2005 to 15%
related health
contracts are
in 2010.
insurance.
assumed to increase.
-
Employees carry two
-
Both direct and
thirds of increased
indirect costs of
wage costs due to
skilled and highly
HIV/AIDS.
skilled employees
-
Employees pass on
would increase from
half of the increase in
5% in 2005 to 10% in
wage costs that they
2010.
have to bear to the
-
60% of the increase
consumer in the form
in direct costs was
of higher prices; this
related to increases
results in a higher
in the costs of
PPI than in the no-
medical benefits.
AIDS case.
-
Companies carry
-
The remaining half of
50% of the direct and
the increase in wage
all of the indirect cost
costs will be born by
increases resulting
firms in the form of
from HIV/AIDS.
lower operating
-
Companies pass half
surpluses.
of these cost
increases onto
consumers in the
form of prices, while
the remaining 50%
will be absorbed
through a reduction
in operating
surpluses.

---

9
D. Government
-
Higher government
−
Higher government
-
Lower population will
-
Government carries
spending
spending on health
expenditure due to
imply a reduction in
50% of direct costs in
and social services;
AIDS.
the overall demand
the public sector.
this will either
-
Public sector spends
for government
-
Expected proportion
displace other
between R3000 and
services other than
of AIDS sufferers
spending or result in
R4500 per AIDS
health services.
who are skilled or
a higher deficit (or a
patient per year.
-
A reduction in the
highly skilled is 11%
lower surplus).
-
Expected annual
demand for
and 14% for 2000 to
-
Health share of total
increases in health
government services
2015.
government
care spending in
will result in the
-
Skilled and highly
spending will rise
excess of R4 billion
decline in
skilled attend private
from 15% in 1997 to
by 2008.
government
health care facilities.
26% in 2010,
-
Higher government
employment set at
-
Unskilled AIDS
representing a 6.9%
spending is financed
50% of the decline in
sufferers attend
average annual
by a higher budget
the non-government
public health care
increase in public
deficit, expenditure
labour force.
facilities, assuming a
health care
switching within the
-
Higher government
75% take-up rate.
expenditure.
health department or
expenditure due to
-
Cost of medical
-
Non-AIDS spending
by sacrificing other
AIDS.
treatment of
stays at a constant
expenditure.
-
Government will
HIV/AIDS is R11 506
proportion of total
carry 50% of the
per patient per year.
absorption.
increased direct

-
Real expenditure on
costs, and
social programs
employees will carry
increased at 2.7%
the other 50%.
per annum.
-
Cost of providing
health care per
AIDS-case is R16
900.
-
75% of all AIDS
victims not employed
in the skilled or
highly skilled sectors
attend public health
care facilities.
-
Public sector
spending on health
care rises from R6
billion in 2005 to
R11.5 billion in 2010.
-
By 2010, additional
welfare spending on
orphans will equal
R2.9 billion,
assuming that 30%
of foster parents
would turn to the
government for
financial assistance.
-
Government will
finance 50% of the
increased
expenditure by
cutting back on other
expenditure.

---

10
E. Household
-
HIV/AIDS affected
-
Lower population
-
As a result of
-
Households finance
spending
households shift their
due to HIV/AIDS will
HIV/AIDS, many
50% of their share of

spending towards
cause an initial
consumers are likely
HIV/AIDS treatment
health related
increase in per capita
to face additional
cost from savings
expenditure.
income of
out-of-pocket health
and the other 50%
-
HIV/AIDS-affected
households.
care spending (not
through decreasing
households save
-
As per capita income
covered by medical
expenditure.
nothing.
increases in the
funds); half of these
-
HIV/AIDS-affected
AIDS scenario,
costs will be financed
households increase
consumption will shift
from savings, and
their share of health
away from non-
the remainder from
service spending
durables towards
cutting back on non-
between 10% and
durable goods and
health care
15% at the expense
services.
expenditure.
of other non-food
-
Increased demand
-
Employees carry
expenditures.
for health services
50% of direct cost
diverts funds and
increases due to
resources away from
HIV/AIDS; they
other expenditure
finance half of this
categories.
increase from
-
25% of higher cost to
personal savings and
the private s ector will
the other half by
be spent on health-
reducing
related services
consumption
-
Demand for health
expenditure.
services could be
-
60% of the direct
over 11% higher by
costs are spent on
2010 than in the no-
healthcare.
AIDS scenario.

3.1
Population and labour supply

High unemployment rates mean that the effect of HIV/AIDS on labour supply in South Africa may
not be that pronounced (Ford et al., 2002). Yet, HIV prevalence rates differ substantially across skill
groups and the epidemic will therefore have a differential impact on labour force growth by skill
category (Arndt and Lewis, 2000, BER, 2001, ING Barings, 2000). In South Africa there is an
inverse relationship between HIV prevalence and skill class, with unskilled and semi-skilled workers
having much higher prevalence rates than their skilled or highly skilled counterparts. Due to this fact
and the current composition of the labour force5, projected losses in the labour force at lower skill
levels far exceed losses at higher skill levels (Russell, 2002), while the epidemic is also likely to
exacerbate the skills shortage in the country (Ford et al., 2002). In aggregate terms, the
macroeconomic models (based on assumptions about prevalence and morbidity and mortality
effects by skill level) all assume that the total labour force or supply will decline and that labour
productivity will decline due to the HIV/AIDS epidemic.

This paper does not aim to evaluate the demographic projections employed in these
macroeconomic models. Yet, it is important to point out that these projections are key to the results
of the modelling and that the demographic projections in turn rely very much on the available HIV
prevalence data. As mentioned above, most of these models employ the demographic projections

5 According to the 1996 census, for example, 62.3% of the South African labour force was semi- or
unskilled, whereas 27.5 and 10.2% respectively were classified as skilled and highly skilled (BER,
2001: 11).

---